LSR3 analysis: effects of TAAR1 agonists in animal models of psychosis

1. Flow of study selection and descriptives

The flow of study selection is shown in Figure 1. Studies included were published between 2011 and 2023. Overall, this analysis includes 15 studies containing 216 comparisons.

Figure 1 - PRISMA flowchart

The table below gives a summary of the included studies, the model and species used, the intervention tested, and the outcome measured. N represents an aggregate of animals contributing to outcomes reported from control and treatment groups, and if the same control group has contributed to more than one experiment, it will be counted twice.

Study	Model	Strain	Comparison	Outcome	N
BEGNI, 2021	Pharmacological	Lister hooded (rat)	SEP-363856 v Vehicle	Cognition	40
“	“	“	“	Locomotor activity	120
CINQUE, 2018	Genetic	Wistar (rat)	RO5203648 v Vehicle	Cognition	32
DEDIC, 2019	Pharmacological	C57BL/6J (mouse)	SEP-363856 v Vehicle	Locomotor activity	48
“	“	“	SEP-363856 v clozapine	Locomotor activity	48
“	“	Sprague-dawley (rat)	SEP-363856 v Vehicle	Social interaction	48
“	“	“	SEP-363856 v clozapine	Social interaction	48
GALLEY, 2012	Pharmacological	Wistar (rat)	RO5073012 v Vehicle	Locomotor activity	48
KOKKINOU, 2021	Pharmacological	C57BL/6 (mouse)	SEP-363856 v Vehicle	Neurobiological outcome	17
KRASAVIN, 2022a	Genetic	Wistar (rat)	LK000764 v Vehicle	Locomotor activity	108
KRASAVIN, 2022a	Pharmacological	Wistar (rat)	LK000764 v Vehicle	Locomotor activity	140
KRASAVIN, 2022b	Genetic	Wistar (rat)	AP163 v Vehicle	Locomotor activity	18
LEO, 2018	Genetic	Wistar (rat)	RO5203648 v Vehicle	Locomotor activity	24
LIANG, 2022	Pharmacological	ICR (mouse)	SEP-363856 & olanzapine v olanzapine	Cognition	48
“	“	“	“	Locomotor activity	16
“	“	“	SEP-363856 v Vehicle	Cognition	192
“	“	“	“	Locomotor activity	96
“	“	“	SEP-363856 v olanzapine	Cognition	48
“	“	“	“	Locomotor activity	16
REVEL, 2011	Genetic	C57BL/6J (mouse)	RO5166017 v Vehicle	Locomotor activity	42
“	Pharmacological	C57BL/6 (mouse)	RO5166017 v Vehicle	Locomotor activity	200
“	“	“	“	Stereotypy	128
“	“	NMRI (mouse)	RO5166017 v Vehicle	Locomotor activity	84
REVEL, 2012a	Genetic	C57Bl/6Jx129Sv/J (mouse)	RO5203648 v Vehicle	Locomotor activity	48
“	Pharmacological	C57BL/6J (mouse)	RO5203648 v Vehicle	Locomotor activity	154
“	“	Wistar (rat)	RO5203648 v Vehicle	Locomotor activity	84
REVEL, 2012b	Pharmacological	C57BL/6J (mouse)	RO5073012 v Vehicle	Locomotor activity	42
REVEL, 2013	Pharmacological	C57BL/6J (mouse)	RO5256390 v Vehicle	Locomotor activity	122
“	“	“	RO5256390 v olanzapine	Locomotor activity	32
“	“	“	RO5263397 & risperidone v risperidone	Locomotor activity	96
“	“	“	RO5263397 v Vehicle	Locomotor activity	184
“	“	“	RO5263397 v olanzapine	Locomotor activity	80
“	“	“	RO5263397 v risperidone	Locomotor activity	96
“	“	Long-evans (rat)	RO5256390 v Vehicle	Cognition	48
“	“	Not”stated (mouse)	RO5256390 v Vehicle	Locomotor activity	80
“	“	“	RO5263397 v Vehicle	Locomotor activity	128
SAARINEN, 2022	Pharmacological	Not stated (mouse)	SEP-363856 v Vehicle	Locomotor activity	56
“	“	“	“	Prepulse inhibition	60
WANG, 2023	Pharmacological	C57BL/6J (mouse)	Compound 50A v Vehicle	Locomotor activity	72
“	“	“	Compound 50B v Vehicle	Locomotor activity	90
“	“	“	Compound 50B v aripiprazole	Locomotor activity	16
“	“	“	Compound 50B v risperidone	Locomotor activity	16
“	“	“	SEP-363856 v Vehicle	Locomotor activity	18
“	“	“	SEP-363856 v aripiprazole	Locomotor activity	16
“	“	“	SEP-363856 v risperidone	Locomotor activity	16
“	“	“	SEP-363856 v risperidone	Locomotor activity	16

References of included studies are located in the appendix. Included studies used 38 unique disease model induction procedures.

1.1 Description of experiment types and methodological approach

Within the literature we identified distinct categories of experiments and the data presented would allow several meta-analytical contrasts to be drawn:

1. TAAR1 agonist vs control. These were experiments investigating the effect of administering a TAAR1 agonist alone, reported in 156 experiments from 15 publications.

2. TAAR1 agonist vs ‘known’ antipsychotic drug. These were experiments investigating the effect of administering a TAAR1 agonist alongside a currently licensed anti-psychotic reported in 27 experiments from 4 publications.

3. Co-treatment with TAAR1 agonist plus know antipsychotic drug v known antipsychotic drug alone, reported in 10 experiments from 2 publications.

4. Effect of TAAR1 antagonism on the effect of TAAR1 agonist v control. These were experiments investigating whether any effect of TAAR1 agonism was unhibited by TAAR1 antagonism. In this iteration of the review, all experiments within this category used genetic approaches to TAAR1 antogonism (that is, they knocked out the gene for the TAAR1 receptor, so any observed drug effect could not be due to actions mediated through the TAAR1 receptor and therefore could not be considered specific drug effects mediated through the TAAR1 receptor.

Each experiment type will be analysed separately. This is because each experiment type uses different control conditions.

In these studies the:

• Control group is a group of animals that is (1) subjected to a psychosis model induction paradigm and (2) administered a control treatment (vehicle) or no treatment

• Intervention group is a group of animals that is (1) subjected to a psychosis model induction paradigm and (2) administered a TAAR1 agonist treatment

• Sham group is a group of animals that is (1) not subjected to a psychosis model induction paradigm and (2) administered a control treatment (vehicle) or no treatment. These data are required to allow a ‘normalised mean difference’ (NMD) effect size to be calculated, given by

  \[ \frac{(\text{$\bar{\mu}_C - \bar{\mu}_T$})} {(\text{$\bar{\mu}_C - \bar{\mu}_S$)}} \text{ x 100} \]

where \(\bar{\mu}_C\), \(\bar{\mu}_T\), \(\bar{\mu}_S\) are the mean reported scores in the control, treatment, and sham groups respectively.

Outcomes with ≥2 independent effect sizes were considered for meta-analysis. In this iteration of the review, this includes locomotor activity and cognition.

All analyses were conducted allowing for the following hierarchical levels in a random effects model, which accounts for features common to experimental contrasts such as a shared control group:

• Level 1: Rodent strain - effect sizes measured across experiments using the same rodent strain

• Level 2: Study - effect sizes measured from different experiments presented in the same publication

• Level 3: Experiment - effect sizes measured in the same experiment within a study, where often a control group contributes to several effect sizes

The hierarchical grouping may therefore be considered thus: Strains of laboratory animals are included in several Studies, each of which can report one or more Experiments, and each Experiment is comprised of at least two Cohorts which are considered identical except for differing in the experimental manipulation (the Intervention) or not being exposed to the disease modelling procedures (a Sham cohort, these only being used to provide a baseline for outcome measures to allow Normalised Mean Difference meta-analysis). An Experiment can include several experimental contrasts, for instance where different doses of drugs are compared to the same control group.

For some experimental contrasts, more than one locomotor or cognitive outcome - for instance both horizontal and vertical climbing activity - was measured in the same cohort of animals. Alternatively, some publications used the same drug doses with the same outcome measures in different experiments. For these reasons, some of the forest plots may appear to include ‘duplicate’ Study - Drug - Dose combinations with different outcomes. For the later, these are accounted for in the heirarchical analysis, but for the former there were insufficient levels of the different locomotor or cognitive outcome measures to allow for hierachical analysis and so this was not performed.

2 TAAR1 Agonists v Control

15 studies (156 comparisons) investigated the effects of TAAR1 Agonist versus Control. The number of studies and individual effect sizes for each outcome were:

• Locomotor activity*: 13 studies and 125 comparisons in 8 strains

• Prepulse inhibition*: 1 studies and 3 comparisons in 1 strain

• Cognitive function: 4 studies and 19 comparisons in 4 strains

• Social interaction: 1 studies and 3 comparisons in 1 strain

• Stereotypy: 1 studies and 5 comparisons in 1 strain

* These outcomes were identified in the study protocol as primary outcomes of interest.

Only one publication reported each of prepulse inhibition (a primary outcome), social interaction, and stereotypy, and so these outcomes are not analysed further.

2.1 Outcome 1: Locomotor Activity

2.1.1 Risks of bias

Figure 2.1.1 shows the risk of bias summary for studies investigating the effect of administering a TAAR1 agonist on locomotor activity in animals. The risk of bias assessment was performed using the SyRCLE’s RoB tool.

Figure 2.1.1 - Traffic light plot of the risk of bias for locomotor activity

2.1.2 Reporting completeness

Figure 2.1.2 shows the reporting completeness summary for studies investigating the effect of administering a TAAR1 agonist on locomotor activity in animals. The reporting completeness assessment was performed using the ARRIVE guidelines.

Figure 2.1.2 - Traffic light plot of the reporting completeness for locomotor activity

2.1.3 Meta-analysis

The effect of administering a TAAR1 agonist on locomotor activity in animals using SMD as the effect size is shown in Figure 2.1.3. The pooled estimate for SMD across all individual comparisons is displayed as a diamond shape at the bottom of the plot. Dotted lines indicate the prediction interval of the pooled estimate.

Figure 2.1.3 - Forest plot of locomotor activity for TAAR1 Agonist vs control

For TAAR1 Agonist v Control, TAAR1 interventions had a pooled effect on locomotor activity of SMD = 1.032 (95% CI: 0.751 to 1.313, with a prediction interval of 0.008 to 2.056).

125 experimental comparisons were reported in 39 experiments reported from 13 publications and involving 8 different animal strains. The between strain variance was 0.017, the between study variance 0.016 and the within-experiment variance 0.14.

2.1.4 Subgroup analyses and meta-regressions

For each outcome, the covariates of interest for subgroup analysesand meta-regressions were:

• Sex

• Method of disease induction

• Route of intervention administration

• Whether the intervention was prophylactic or therapeutic (i.e. administered before or after disease model induction)

• Duration of treatment period

• The intervention administered

• The efficacy of the drug (i.e. whether the drug is a partial or full agonist)

• The selectivity of the drug

• Potency of the intervention

• Dose of intervention

We also conducted subgroup analyses using (1) SyRCLE Risk of Bias and (2) ARRIVE reporting completeness assessment scores as covariates to evaluate their influence on effect size estimates. These were not specified in the study protocol, but evaluation of risk of bias is required for the Summary of Evidence table, and no studies were considered at low risk of bias or high reporting completeness to allow such a sensitivity analysis

The significance (p value) reported is that for a test of whether the moderators are significantly different one from another, rather than whether the effect is significantly different from 0. Note that in this iteration of the review, insufficient data was available to conduct subgroup analysis for the following variables, as only one subgroup level was present: duration of treatment period

Sex

Figure 2.1.4.1 displays the estimates for the pooled SMD’s when comparisons are stratified by sex of the animal. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by sex, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.1 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by sex

The p-value for the association between the sex of animal groups used and outcome reported was 0.671. The between-strain variance was 0, the between-study variance 0.052, and the within-experiment variance 0.125.

Category of disease induction

Figure 2.1.4.2 displays the estimates for the pooled SMD’s when comparisons are stratified by the category of disease induction. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by category of disease induction, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.2 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by category of disease induction

The p-value for the association between whether genetic or pharmacological models were used and outcome reported was 0.709. The between-strain variance was 0.027, the between-study variance 0.016, and the within-experiment variance 0.145.

Route of intervention administration

Figure 2.1.4.3 displays the estimates for the pooled SMD’s when comparisons are stratified by the route of intervention administration. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by route of intervention administration, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.3 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by route of intervention administration

The p-value for the association between the route of intervention administration and outcome reported was 0.608. The between-strain variance was 0.032, the between-study variance 0.002, and the within-experiment variance 0.15.

Prophylactic or therapeutic intervention

Figure 2.1.4.4 displays the estimates for the pooled SMD’s when comparisons are stratified by whether the intervention was administered prophylactically or therapeutically. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by whether the intervention was administered prophylactically or therapeutically, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.4 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by intervention type

The p-value for the association between whether the intervention was administered prophylactically or therapeutically and outcome reported was 0.608. The between-strain variance was 0.032, the between-study variance 0.002, and the within-experiment variance 0.15.

Duration of treatment period

In this iteration of the review, all relevant comparisons administered the TAAR1 agonist for < 1 week. Therefore, no subgroup analyses were conducted for this variable.

---

The intervention administered

Figure 2.1.4.6 displays the estimates for the pooled SMD’s when comparisons are stratified by the intervention administered. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by the intervention administered, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.6 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by intervention administered

The p-value for the association between the intervention and outcome reported was 0.448. The between-strain variance was 0.063, the between-study variance 0, and the within-experiment variance 0.117.

The efficacy of the drug (i.e. whether the drug is a partial or full agonist)

Figure 2.1.4.7 displays the estimates for the pooled SMD’s when comparisons are stratified by the action/efficacy of the intervention administered. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by intervention efficacy, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.7 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by efficacy of the drug

The p-value for the association between whether the drug was a full or partial agonist and outcome reported was 0.318. The between-strain variance was 0.024, the between-study variance 0.03, and the within-experiment variance 0.129.

The selectivity of the drug

Figure 2.1.4.8 displays the estimates for the pooled SMD’s when comparisons are stratified by the selectivity of the intervention administered. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by intervention selectivity, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.8 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by selectivity of the drug

The p-value for the association between whether the drug was highly selective, or also manifests 5-HT1A effects, and outcome reported was 0.297. The between-strain variance was 0, the between-study variance 0.016, and the within-experiment variance 0.146.

Potency of intervention

The pEC50 value of each drug was used to measure potency. The pEC50 value is the negative logarithm (to base 10) of the EC50 value. Higher pEC50 values indicate higher potency (as they indicate a lower EC50). Figure 2.1.4.9 displays a visualisation of the meta-regression using the pEC50 value as an explanatory variable. Dashed lines represent the 95% confidence interval of the regression line. The dotted lines represent the 95% prediction interval. Raw data are plotted with ‘bubble’ size adjusted according to effect size precision.

Figure 2.1.4.9 - Meta-regression of TAAR1 agonist vs control on locomotor activity by potency of intervention

The estimate for \(\beta\) was -0.021 (p = 0.907). The between-strain variance was 0, the between-study variance 0.041, and the within-experiment variance 0.142.

Dose of intervention

In this iteration of the review, the TAAR1 agonists tested against control for their effect on locomotor activity were: RO5203648, RO5263397, SEP-363856, RO5166017, LK000764, RO5256390, Compound 50B, Compound 50A, RO5073012 and AP163. The dashed lines in the plot represent the 95% confidence interval of the regression line and the dotted lines represent the 95% prediction interval. Raw data are plotted with point size adjusted according to effect size precision.

RO5203648: There were 21 comparisons from 2 publication(s).

RO5263397: There were 21 comparisons from 1 publication(s).

SEP-363856 (Ultaront): There were 19 comparisons from 5 publication(s).

RO5166017: There were 18 comparisons from 1 publication(s).

LK000764: There were 16 comparisons from 1 publication(s).

RO5256390: There were 14 comparisons from 1 publication(s).

Compound 50B: There were 5 comparisons from 1 publication(s).

Compound 50A: There were 4 comparisons from 1 publication(s).

RO5073012: There were 4 comparisons from 2 publication(s).

AP163: There were 3 comparisons from 1 publication(s).

Standardised dose

We then sought evidence of a dose response relationship across all drugs. To do this, we conducted meta-regression using a constructed variable, the ‘standardised dose’. The EC50 of a drug is the molar concentration at which 50% of the maximal response occurs. While the drug concentrations achieved at the receptor are unknown, we can approximate this from the dose given (expressed as g/kg), and the molar mass of the drug (g/mol). This relies on an approximation that the drug is equally distributed throughout the animal, and so does not take into account for example first pass metabolism for orally administered drugs, blood brain barrier solubility or differential accumulation in fatty tissues. As such, it should be interpreted with extreme caution; but does provide allow some imputation of whether, across all drugs, there is a dose-response effect. On this measure, a standardised dose of 0 would reflect 50% of maximum effect and a standardised dose of 1 would reflect around 80% of maximum effect

The standardised dose was calculated as the logarithm of the dose of the intervention (in g/kg) divided by the product of the intervention’s EC50 (in moles) and the Molar mass of the drug (in g/mol):

\[ \log\frac{(\text{Dose of Intervention (g/kg)})}{(\text{Molar Mass (g/mol)}) \times ({\text{EC50 (mol)}})} \]

This is a simplified approximation based on the reasoning that if drug actions are mediated through the TAAR1 receptor, and drug efficacy is reflected in the respective EC50 values, then in principal drugs should exhibit similar effects when acting at their respective EC50.

The actual concentration of a drug at the receptor site is influenced by several variables, including dosage, administration route, elimination half-life, and first-pass metabolism (in case of oral administration). Incorporating all these factors accurately would necessitate a detailed pharmacokinetic model, which falls outside the scope of this review. Consequently, we assume uniformity across experiments in terms of (i) volume of distribution, (ii) first-pass metabolism, (iii) blood-brain barrier permeability, and (iv) experimental design, especially regarding the timing of peak drug concentration (where we assume that experiments were designed to be done at a time when the drug was near peak concentration).

Figure 2.1.4.10 provides a visualisation of the meta-regression analysis relationship between standardised doses of TAAR1 agonists and the Standardized Mean Difference (SMD) change in Locomotor activity. As before, dashed lines represent the 95% confidence interval of the regression line and dotted lines represent the 95% prediction interval. Raw data are plotted with point size adjusted according to effect size precision.

Figure 2.1.4.10 - Meta regression of standardised dose for TAAR1 agonist vs control on locomotor activity

The estimate for the change in effect per log unit change in standardised dose was 0.2 (p < 0.001). The between-strain variance was 0, the between-study variance 0.438, and the within-experiment variance 0.095.

SyRCLE RoB assessment

Figure 2.1.4.11 displays the estimates for the pooled SMD’s when comparisons are stratified by how many of the SyRCLE risk of bias assessment criteria (of which there are 10) that the experiment met. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by SyRCLE Risk of Bias, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.11 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by SyRCLE RoB criteria met

The p-value for the association between SyRCLE Risks of Bias reporting and outcome reported was 0.215. The between-strain variance was 0.017, the between-study variance 0, and the within-experiment variance 0.141.

Alternative SyRCLE RoB assessment

Figure 2.1.4.12 displays the estimates for the pooled SMD’s when comparisons are stratified by whether of not any of the SyRCLE Risk of bias domains were rated as low risk of bias. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by SyRCLE Risk of Bias, is displayed as a diamond shape at the bottom of the plot.

Figure 2.1.4.12 - Subgroup analysis of TAAR1 agonist vs control on locomotor activity by alternative SyRCLE RoB assessment

The p-value for the association between low SyRCLE Risks of Bias reporting and outcome reported was 0.229. The between-strain variance was 0.001, the between-study variance 0, and the within-experiment variance 0.162.

ARRIVE reporting completeness guidelines

Experiments were categorised based on the number of ARRIVE guidelines items (of which there are 23) met.

Figure 2.1.4.13 displays a visualisation of the meta-regression using the number of ARRIVE items met as an explanatory variable. Dashed lines represent the 95% confidence interval of the regression line. The dotted lines represent the 95% prediction interval. Raw data are plotted with ‘bubble’ size adjusted according to effect size precision.

Figure 2.1.4.13 - Meta-regression of number of ARRIVE items met for TAAR1 agonist vs control on locomotor activity

The estimate for \(\beta\) was -0.004 (p = 0.923). The between-strain variance was 0, the between-study variance 0.041, and the within-experiment variance 0.141.

Heterogeneity explained by covariates (TAAR1 Agonist vs Control on locomotor activity)

The table below shows which of the covariates, if any, explain some of the heterogeneity observed in the effect sizes of the effect of TAAR1 agonists on locomotor activity. We present marginal R², which measures the proportion of variance explained by including moderators in the model (the % change in the between-studies variance when the covariate is included in the model). The coefficients are derived form an RMA model fitted with an intercept (and so represent, for each category, the point estimate and 95% CIs of the effect in that category).

Moderator	Category	\(\beta\)	95% CI	Marginal R2 (%)
Overall effect	-	1.032	0.751 to 1.313	-
Sex	-	-	-	8.6%
-	Female	0.847	0.011 to 1.683	-
-	Male	1.185	0.818 to 1.551	-
-	Mixed	0.819	0.169 to 1.469	-
-	Not reported	0.964	0.604 to 1.323	-
Category of disease model induction	-	-	-	0.9%
-	Genetic	1.139	0.516 to 1.891	-
-	Pharmacological	1.018	0.749 to 1.288	-
Administration route	-	-	-	2.8%
-	Intraperitoneal	0.866	0.109 to 1.622	-
-	Oral	1.06	0.78 to 1.34	-
Prophylactic or therapeutic intervention	-	-	-	2.8%
-	Prophylactic	1.06	0.78 to 1.34	-
-	Therapeutic	0.866	0.109 to 1.622	-
Intervention administered	-	-	-	41.6%
-	AP163	1.319	-0.677 to 3.315	-
-	Compound 50A	0.665	-0.459 to 1.79	-
-	Compound 50B	1.109	-2.086 to 1.666	-
-	LK000764	0.407	-0.575 to 1.388	-
-	RO5073012	0.6	-0.293 to 1.492	-
-	RO5166017	1.324	0.685 to 1.963	-
-	RO5203648	1.006	0.406 to 1.607	-
-	RO5256390	1.694	0.976 to 2.412	-
-	RO5263397	0.929	0.339 to 1.518	-
-	SEP-363856 (Ultaront)	0.936	0.476 to 1.397	-
Drug efficacy	-	-	-	5.3%
-	Full agonist	1.108	0.809 to 1.407	-
-	Partial agonist	0.906	0.561 to 1.25	-
Drug selectivity	-	-	-	18.7%
-	High	1.14	0.864 to 1.417	-
-	Low	0.945	0.534 to 1.356	-
-	Unclear	0.637	0.014 to 1.261	-
Drug potency	per log unit	-0.021	-0.382 to 0.339	0.1%
Standardised drug dose	per log unit	0.2	0.151 to 0.248	34.2%
Risk of Bias	-	-	-	22.1%
-	0 criteria met	1.105	0.842 to 1.368	-
-	1 criteria met	0.593	-0.014 to 1.201	-
-	2 criteria met	1.545	0.237 to 2.854	-
Reporting completeness	-	-	-	3.1%
-	4-7 criteria met	1.471	-0.424 to 3.367	-
-	8-11 criteria met	1.07	0.69 to 1.45	-
-	12-15 criteria met	0.972	0.55 to 1.393	-
-	16-19 criteria met	1.096	-0.119 to 2.311	-

2.1.5 Sensitivity Analyses

We examine the robustness of the findings for the primary outcome by performing the following sensitivity analyses

Imputed 𝞺 values of 0.2 and 0.8

In the previous analyses for the effect of TAAR1 agonists on locomotor activity, we imputed a \(\rho\) value - the imputed within-study correlation between observed effect sizes - of 0.5. Here, we examine the effect of imputing \(\rho\) values of 0.2 and 0.8.

When the \(\rho\) value is assumed to be 0.2, the TAAR1 interventions had a larger effect on locomotor activity of SMD = 1.15 (95% CI: 0.91 to 1.4) with a prediction interval of 0.09 to 2.21).

When the \(\rho\) value is assumed to be 0.8, the TAAR1 interventions had a smaller and more imprecise effect on locomotor activity of SMD = 0.72 (95% CI: 0.07 to 1.38) with a prediction interval of -1.33 to 2.78).

For reference the pooled effect size when rho is assumed to be 0.5 is 1.03 (95% CI: 0.75 to 1.31). Therefore, the effect is very sensitive to imputed within-study correlation between effect sizes.

Normalised Mean Difference (NMD)

For locomotor activity, 96 out of 125 comparisons, i.e. 76.8 % of comparisons, had data available for a Sham group and, therefore, to calculate an NMD.

The effect of administering a TAAR1 agonist on locomotor activity in animals using NMD as the effect size is shown in Figure 2.1.5. The pooled estimate for NMD across all individual comparisons is displayed as a diamond shape at the bottom of the plot. Dotted lines indicate the prediction interval of the pooled estimate.

Figure 2.1.5 - Forest plot of TAAR1 agonist vs control on locomotor activity using NMD

For TAAR1 Agonist v Control, TAAR1 interventions had a pooled effect on locomotor activity of NMD = 56.86 (95% CI: 44.08 to 69.63) with a prediction interval of -13.72 to 127.44). For reference the pooled effect size for SMD was 1.03 (95% CI: 0.75 to 1.31).

96 experimental comparisons were reported in 30 experiments reported from 10 publications and involving 8 different animal strains. Between-strain variance was 0, between-study variance was 0.02, and within-study variance (between experiments) was 861.72.

Robust variance estimator (RVE)

Here, we examine the robustness of results when using a sandwich-type estimator to obtain cluster-robust tests and confidence intervals of the model coefficients. The variance-covariance matrix is estimated using the ‘bias-reduced linearization’ for small-sample adjustment and Strain as a clustering variable.

When using the robust variance estimator, TAAR1 interventions had a pooled effect on locomotor activity of SMD = 1.03 (95% CI: 0.73 to 1.33 with a prediction interval of -0.12 to 2.18). For reference the pooled effect size for SMD was 1.03 (95% CI: 0.75 to 1.31), so the using a robust variance estimator does not substantially change the results.

2.1.6 Reporting bias/small-study effects

Because of the relationship between SMD effect sizes and variance inherent in their calculation, where study size is small the standard approach to seeking evidence of small-study effects (regression based tests including Egger’s regression test for multilevel meta-analysis) can lead to over-estimation of small-study effect (see for instance 10.7554/eLife.24260). To address this we used Egger’s regression test for multilevel meta-analysis, with regression of SMD effect size against 1/√N, where N is the total number of animals involved in an experiment.

Egger regression based on 125 effects of TAAR1 Agonist v Control where Locomotor activity was measured showed a coefficient for small-study effect of 7.93 (95% CI: 3.05 to 12.82; p = 0.002).

2.2 Outcome 2: Cognitive function

2.2.1 Risks of bias

Figure 2.2.1 shows the risk of bias summary for studies investigating the effect of administering a TAAR1 agonist on cognition in animals. The risk of bias assessment was performed using the SyRCLE’s RoB tool.

Figure 2.2.1 - Traffic light plot of the risk of bias for cognitive function

2.2.2 Reporting completeness

Figure 2.2.2 shows the reporting completeness summary for studies investigating the effect of administering a TAAR1 agonist on cognition in animals. The reporting completeness assessment was performed using the ARRIVE guidelines.

Figure 2.2.2b - Traffic light plot of the reporting completeness for cognitive function

2.2.3 Meta-analysis

The effect of administering a TAAR1 agonist on cognitive outcomes in animals using SMD as the effect size is shown in Figure 2.2.3. The pooled estimate for SMD across all individual comparisons is displayed as a diamond shape at the bottom of the plot. Dotted lines indicate the prediction interval of the pooled estimate.

Figure 2.2.3 - Forest plot of cognitive function for TAAR1 Agonist vs control

For TAAR1 Agonist v Control, TAAR1 interventions had a pooled effect on cognitive outcomes of SMD =0.8 (95% CI: -0.301 to 1.9, with a prediction interval of-1.609 to 3.208).

19 experimental comparisons were reported in 0 experiments reported from 4 publications and involving 5 different animal strains. The between strain variance was 0.6, the between study variance NA and the within-experiment variance NA.

2.2.4 Subgroup analyses and meta-regressions

Sex

Figure 2.2.4.1 displays the estimates for the pooled SMD’s when comparisons are stratified by sex of the animal. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by sex, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.1 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by sex

The p-value for the association between the sex of animal groups used and outcome reported was 0.46. The between-strain variance was 0.24, the between-study variance 0.24, and the within-experiment variance 0.

Category of disease induction

Figure 2.2.4.2 displays the estimates for the pooled SMD’s when comparisons are stratified by the category of disease induction. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by category of disease induction, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.2 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by category of disease induction

The p-value for the association between whether genetic or pharmacological models were used and outcome reported was 0.27. The between-strain variance was 0.27, the between-study variance 0.27, and the within-experiment variance 0.

Route of intervention administration

Figure 2.2.4.3 displays the estimates for the pooled SMD’s when comparisons are stratified by the administration route of the intervention. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by administration route of the intervention, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.3 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by route of intervention administration

The p-value for the association between whether genetic or pharmacological models were used and outcome reported was 0.27. The between-strain variance was 0.27, the between-study variance 0.27, and the within-experiment variance 0.

Prophylactic or therapeutic intervention

In this iteration of the review, all relevant comparisons administered the TAAR1 agonist after induction of the disease model. Therefore, no subgroup analyses were conducted for this variable.

Duration of treatment period

In this iteration of the review, all relevant comparisons administered the TAAR1 agonist for < 1 week. Therefore, no subgroup analyses were conducted for this variable.

The intervention administered

Figure 2.2.4.4 displays the estimates for the pooled SMD’s when comparisons are stratified by the intervention administered. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by the intervention administered, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.4 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by intervention administered

The p-value for the association between the intervention and outcome reported was 0.64. The between-strain variance was 0.58, the between-study variance 0.58, and the within-experiment variance 0.

The efficacy of the drug (i.e. whether the drug is a partial or full agonist)

In this iteration of the review, all relevant comparisons administered the TAAR1 agonists with partial agonist activity. Therefore, no subgroup analyses were conducted for this variable.

The selectivity of the drug

Figure 2.2.4.5 displays the estimates for the pooled SMD’s when comparisons are stratified by the selectivity of the intervention administered. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by intervention selectivity, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.5 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by selectivity of the drug

The p-value for the association between whether the drug was highly selective, or also manifests 5-HT1A effects, was 0.38. The between-strain variance was 0.41, the between-study variance 0.41, and the within-experiment variance 0.

Potency of interventions

The pEC50 value of each drug was used to measure potency. The pEC50 value is the negative logarithm (to base 10) of the EC50 value. Higher pEC50 values indicate higher potency (as they indicate a lower EC50). Figure 2.2.4.6 displays a visualisation of the meta-regression using the pEC50 value as an explanatory variable. Dashed lines represent the 95% confidence interval of the regression line. The dotted lines represent the 95% prediction interval. Raw data are plotted with ‘bubble’ size adjusted according to effect size precision.

Figure 2.2.4.6 - Meta-regression of TAAR1 agonist vs control on cognitive function by potency of the interventions

The estimate for \(\beta\) was -1.19 (p = 0.47). The between-strain variance was 0.5, the between-study variance 0.5, and the within-experiment variance 0.

Dose of intervention

In this iteration of the review, the TAAR1 agonists tested against control for their effect on cognition were; SEP-363856, RO5256390 and RO5203648. The dashed lines in the plot represent the 95% confidence interval of the regression line and the dotted lines represent the 95% prediction interval. Raw data are plotted with point size adjusted according to effect size precision.

RO5203648: There were 2 comparisons from 1 publication(s).

RO5263397: There were 0 comparisons from 0 publication(s).

SEP-363856 (Ultaront): There were 14 comparisons from 2 publication(s).

RO5166017: There were 0 comparisons from 0 publication(s).

LK000764: There were 0 comparisons from 0 publication(s).

RO5256390: There were 3 comparisons from 1 publication(s).

Compound 50B: There were 0 comparisons from 0 publication(s).

Compound 50A: There were 0 comparisons from 0 publication(s).

RO5073012: There were 0 comparisons from 0 publication(s).

AP163: There were 0 comparisons from 0 publication(s).

Standardised dose

We then sought evidence of a dose response relationship across all drugs using the approach described for locomotor activity.

Figure 2.2.4.7 provides a visualisation of the meta-regression analysis relationship between standardised doses of TAAR1 agonists and the Standardized Mean Difference (SMD) change in cognition. As before, dashed lines represent the 95% confidence interval of the regression line and dotted lines represent the 95% prediction interval. Raw data are plotted with point size adjusted according to effect size precision.

Figure 2.2.4.7 - Meta regression of standardised dose for TAAR1 agonist vs control on cognitive function

The estimate for the change in effect per log unit change in standardised dose was -0.21 (p = 0.052). The between-strain variance was 0.55, the between-study variance 0.55, and the within-experiment variance 0.

SyRCLE RoB assessment

Figure 2.2.4.8 displays the estimates for the pooled SMD’s when comparisons are stratified by how many of the SyRCLE risk of bias assessment criteria (of which there are 10) that the experiment met. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by SyRCLE RoB assessment, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.8 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by SyRCLE RoB criteria met

The p-value for the association between SyRCLE Risks of Bias reporting and outcome reported was 0.13. The between-strain variance was 0.09, the between-study variance 0.09, and the within-experiment variance 0.

Alternative SyRCLE RoB assessment

Figure 2.2.4.9 displays the estimates for the pooled SMD’s when comparisons are stratified by whether of not any of the SyRCLE Risk of bias domains were rated as low risk of bias. Whiskers indicate the 95% confidence interval of each estimate. The overall pooled SMD, not stratified by SyRCLE Risk of Bias, is displayed as a diamond shape at the bottom of the plot.

Figure 2.2.4.9 - Subgroup analysis of TAAR1 agonist vs control on cognitive function by alternative SyRCLE RoB assessment

The p-value for the association between low SyRCLE Risks of Bias reporting and outcome reported was 0.13. The between-strain variance was 0.09, the between-study variance 0.09, and the within-experiment variance 0.

ARRIVE reporting completeness guidelines

Experiments were categorised based on the number of ARRIVE guidelines items (of which there are 23) met.

Figure 2.2.4.10 displays a visualisation of the meta-regression using the number of ARRIVE items met as an explanatory variable. Dashed lines represent the 95% confidence interval of the regression line. The dotted lines represent the 95% prediction interval. Raw data are plotted with ‘bubble’ size adjusted according to effect size precision.

Figure 2.2.4.10 - Meta-regression of number of ARRIVE items met for TAAR1 agonist vs control on cognitive function

The estimate for \(\beta\) was -0.01 (p = 0.97). The between-strain variance was 0.72, the between-study variance 0.72, and the within-experiment variance 0.

Heterogeneity explained by covariates (TAAR1 Agonist vs Control on cognitive function)

The table below shows which of the covariates, if any, explain some of the heterogeneity observed in the effect sizes of the effect of TAAR1 agonists on cognition. We present marginal R², which measures the proportion of variance explained by including moderators in the model.

Moderator	Category	\(\beta\)	95% CI	Marginal R2 (%)
Overall effect	-	0.8	-0.3 to 1.9	-
Sex	-	-	-	40.1%
-	Female	2.28	-8.68 to 13.24	-
-	Male	0.28	-7.11 to 7.66	-
-	Mixed male and female	0.65	-8.95 to 10.25	-
Category of disease model induction	-	-	-	29.3%
-	Genetic	-0.31	-4 to 3.38	-
-	Pharmacological	1.2	-0.9 to 3.29	-
Administration route	-	-	-	29.3%
-	Intraperitoneal	-0.31	-4 to 3.38	-
-	Oral	1.2	-0.9 to 3.29	-
Intervention administered	-	-	-	20.9%
-	RO5203648	-0.31	-15.03 to 14.41	-
-	RO5256390	0.85	-13.82 to 15.53	-
-	SEP-363856 (Ultaront)	1.41	-8.89 to 11.71	-
Drug selectivity	-	-	-	23.7%
-	High	0.27	-2.79 to 3.34	-
-	Low	1.39	-1.63 to 4.42	-
Drug potency	per log unit	-1.19	-7.05 to 4.67	16.1%
Standardised dose	per log unit	-0.21	-0.42 to 0	9%
Risk of Bias	-	-	-	65.6%
-	0 criteria met	0.43	-0.97 to 1.84	-
-	1 criteria met	2.28	-0.55 to 5.1	-
Reporting completeness	-	-	-	0.6%
-	8-11 criteria met	0.85	-22.24 to 23.95	-
-	12-15 criteria met	0.97	-15.46 to 17.4	-
-	16-19 criteria met	0.65	-22.08 to 23.39	-

2.2.5. Sensitivity Analyses

Imputed 𝞺 values of 0.2 and 0.8

In the previous analyses for the effect of TAAR1 agonists on cognition, we imputed a \(\rho\) value of 0.5. Here, we examine the effect of imputing \(\rho\) values of 0.2 and 0.8.

When the \(\rho\) value is assumed to be 0.2, the TAAR1 interventions had a pooled effect on cognition of SMD = 0.84 (95% CI: -0.27 to 1.95) with a prediction interval of -1.69 to 3.37).

When the \(\rho\) value is assumed to be 0.8, the TAAR1 interventions had a pooled effect on cognition of SMD = 0.65 (95% CI: -0.39 to 1.69) with a prediction interval of -1.5 to 2.81).

For reference the pooled effect size when rho is assumed to be 0.5 is 0.8 (95% CI: -0.3 to 1.9).

Normalised Mean Difference (NMD)

For cognition, 19 out of 19 comparisons, i.e. 100 % of comparisons, had data available for a Sham group and, therefore, to calculate an NMD.

The effect of administering a TAAR1 agonist on cognition in animals using NMD as the effect size is shown in Figure 2.2.5. The pooled estimate for NMD across all individual comparisons is displayed as a diamond shape at the bottom of the plot. Dotted lines indicate the prediction interval of the pooled estimate.

Figure 2.2.5 - Forest plot of TAAR1 agonist vs control on cognitive function using NMD

For TAAR1 Agonist v Control, TAAR1 interventions had a pooled effect on cognition of NMD = 43.1 (95% CI: -38.01 to 124.2) with a prediction interval of -137.84 to 224.03. For reference the pooled effect size for SMD was 0.8 (95% CI: -0.3 to 1.9).

19 experimental comparisons were reported in 5 experiments reported from 4 publications and involving 4 different animal strains. Between-strain variance was 1289.26, between-study variance was 1289.26, and within-study variance (between experiments) was 4.46.

Robust variance estimator (RVE)

Here, we examine the robustness of results when using a sandwich-type estimator to obtain cluster-robust tests and confidence intervals of the model coefficients. The variance-covariance matrix is estimated using the ‘bias-reduced linearization’ for small-sample adjustment and Strain as a clustering variable.

When using the robust variance estimator, TAAR1 interventions had a pooled effect on cognition of SMD = 0.8 (95% CI: -0.58 to 2.18 with a prediction interval of -2.21 to 3.81). For reference the pooled effect size for SMD was 0.8 (95% CI: -0.3 to 1.9), so the using a robust variance estimator does not substantially change the results.

2.2.6 Reporting bias/small-study effects

Because of the relationship between SMD effect sizes and variance inherent in their calculation, where study size is small the standard approach to seeking evidence of small-study effects (regression based tests including Egger’s regression test for multilevel meta-analysis) can lead to over-estimation of small-study effect (see for instance 10.7554/eLife.24260). To address this we used Egger’s regression test for multilevel meta-analysis, with regression of SMD effect size against 1/√N, where N is the total number of animals involved in an experiment.

Egger regression based on 19 effects of TAAR1 Agonist v Control where Locomotor activity was measured showed a coefficient for a small study effect of -68.13 (95% CI: -158.3 to 22.04; p = 0.083).

3 TAAR1 Agonist v known antipsychotic drug

3.1 Outcome 1: Locomotor activity

In TAAR1 Agonist v known antipsychotic drug studies, the effect of administering a TAAR1 agonist on Locomotor activity in animals using SMD as the effect size is shown in Figure 3.1. The pooled estimate for SMD across all individual comparisons is displayed as a diamond shape at the bottom of the plot. Dotted lines indicate the prediction interval of the pooled estimate.

Figure 3.1 - Forest plot of Locomotor activity for TAAR1 Agonist vs known antipychotic drug

For TAAR1 Agonist v known antipsychotic drug comparisons, TAAR1 interventions had a pooled effect on locomotor activity of SMD =-0.622 (95% CI: -1.324 to 0.08, with a prediction interval of-2.272 to 1.029).

21 experimental comparisons were reported in 0 experiments reported from 4 publications and involving 7 different animal strains. The between strain variance was 0.37, the between study variance NA and the within-experiment variance NA.

3.2 Outcome 2: Cognitive function

Multilevel analysis is only performed if there are 5 levels or more for at least one of Strain, Study and Experiment, and that is not the case here. We provide a conventional univariate random effects model to illustrate the data

---

4 Co-treatment with TAAR1 agonist plus known antipsychotic drug v known antipsychotic drug alone

4.1 Outcome 1: Locomotor activity

Multilevel analysis is only performed if there are 5 levels or more for at least one of Strain, Study and Experiment, and that is not the case here. We provide a conventional univariate random effects model to illustrate the data

---

4.2 Outcome 2: Cognitive function

Multilevel analysis is only performed if there are 5 levels or more for at least one of Strain, Study and Experiment, and that is not the case here. We provide a conventional univariate random effects model to illustrate the data

---

5 Effect of TAAR1 agonists in TAAR1 receptor knockout animals

5.1 Outcome 1: Locomotor activity

Multilevel analysis is only performed if there are 5 levels or more for at least one of Strain, Study and Experiment, and that is not the case here. We provide a conventional univariate random effects model to illustrate the data

---

5.2 Outcome 2: Cognitive function

No studies reported cognitive outcomes in TAAR1 knockout animals

---

6. Proportion of animals not progressing to outcome measurement and adverse effects of treatment

2.58% of 1085 animals in Control cohorts and 3.32% of 1085 animals in Intervention cohorts ‘dropped out’ between allocation to group and outcome measurement. Given that 184 of 213 interventions (86.38%) were administered as a single dose, treatment emergent adverse effects likely to lead to withdrawal of an animal from the study would be unusual, and technical failure or attrition is more likely. This analysis is based on full reporting of animals excluded from analyses, and it may be that group sizes were specified ‘after the event’, or that there was unreported replacement of animals excluded during the experiment, so these data should be interpreted with caution.

7. Summary of the evidence

7.1 TAAR1 agonists versus control

Outcome	Timepoint	Summary of the association	Within-study biases	Across-studies biases	Indirectness	Other biases
Locomotor activity	immediate	125 experimental comparisons from 39 experiments in 13 publications involving 8 animal strains and reporting data from 1945 animals; SMD = 1.032 (95% CI: 0.751 to 1.313); Dose effects: there was no significant relationship between standardised dose and outcome	Most studies at unclear risk of bias (SyRCLE); the median number of ARRIVE items reported was 13 (of 23)	Moderate risk: no studies preregistered their analyses; significant evidence for small study effects	see explanatory table	No other risks identified
Cognition	within 1 week	19 experimental comparisons from 5 experiments in 4 publications involving 4 animal strains and reporting data from 312 animals; SMD =0.8 (95% CI: -0.301 to 1.9) (no significant effect); Dose effects: there was no significant relationship between standardised dose and outcome.	Most studies at unclear risk of bias (SyRCLE); the median number of ARRIVE items reported was 15 (of 23)	Moderate risk: no studies preregistered their analyses; no significant evidence for small study effects	see explanatory table	No other risks identified

7.2 TAAR1 agonists versus conventional antipsychotic drugs

Outcome	Timepoint	Summary of the association	Within-study biases	Across-studies biases	Indirectness	Other biases
Locomotor activity	immediate	21 experimental comparisons from 39 experiments in 4 publications involving 7 animal strains and reporting data from 336 animals; SMD =-0.622 (95% CI: -1.324 to 0.08) (non significantly favours conventional antipsychotics); Dose effects: insufficient data	Most studies at unclear risk of bias (SyRCLE); the median number of ARRIVE items reported was 13 (of 23)	Single study	see explanatory table	No other risks identified
Cognition	within 1 week	3 experimental comparisons from 1 experiment in 1 publications involving 1 animal strains and reporting data from 36 animals; insufficient data for further analysis	Most studies at unclear risk of bias (SyRCLE); the median number of ARRIVE items reported was 15 (of 23)	Single study	see explanatory table	No other risks identified

The evaluation of indirectness of evidence is available in the Appendix.

8. Software used

We used R version 4.3.1 (R Core Team 2023) and the following R packages: devtools v. 2.4.5 (Wickham et al. 2022), dosresmeta v. 2.0.1 (Crippa and Orsini 2016), gtools v. 3.9.4 (Bolker, Warnes, and Lumley 2022), Hmisc v. 5.1.1 (Harrell Jr 2023a), kableExtra v. 1.3.9.9001 (Zhu 2023), knitr v. 1.45 (Xie 2014, 2015, 2023), Matrix v. 1.6.1.1 (Bates, Maechler, and Jagan 2023), meta v. 6.5.0 (Balduzzi, Rücker, and Schwarzer 2019), metadat v. 1.2.0 (White et al. 2022), metafor v. 4.4.0 (Viechtbauer 2010), mvmeta v. 1.0.3 (Gasparrini, Armstrong, and Kenward 2012), numDeriv v. 2016.8.1.1 (Gilbert and Varadhan 2019), orchaRd v. 2.0 (Nakagawa et al. 2023), patchwork v. 1.1.3 (Pedersen 2023), PRISMA2020 v. 1.1.1 (Haddaway et al. 2022), rje v. 1.12.1 (Evans 2022), rms v. 6.7.1 (Harrell Jr 2023b), robvis v. 0.3.0.900 (McGuinness and Higgins 2020), tidyverse v. 2.0.0 (Wickham et al. 2019), usethis v. 2.2.2 (Wickham et al. 2023), xtable v. 1.8.4 (Dahl et al. 2019).

Appendix

Evaluation of indirectness of evidence (based on criteria in document “Assessing the certainty of evidence in animal studies”) for the studies included in the review

		Homological validity	Ontopathogenic validity	Triggering validity	Mechanistic validity	Induction validity	Remission validity	Biomarker validity	Ethological validity
Species and strain	Rat	We could find no evidence that the rat behavioural repertoire is closer to human than is the mouse	-	-		-	-	-	-
Mouse	-	-		-	-	-	-
Model Induction	Models using genetic induction – the DAT KO model	Polymorphisms in human SLC6A3 DAT gene reportedly associated with schizophrenia in some populations	-	Indirect DAT inhibitors such as methamphetamine can induce psychosis in humans.	The effect of indirect DAT inhibitors is thought to be mediated through TAAR1 agonism		-	-	-
Pharmacological psychostimulant models cocaine, amphetamine etc NMDA models phencyclidine PCP, MK801	-	No	MK801, ketamine, PCP and amphetamine induce psychosis and exacerbate symptoms in humans	Psychostimulant models induce mesolimbic dopamine dysregulation	1.The chronic PCP model has been associated with reduced brain volume; 2.EEG changes induced by amphetamine, PCP and MK801 are seen in human disease	In vivo EEG changes induced by amphetamine, PCP and MK801 are responsive to treatment	The chronic PCP model has been associated with reduced brain volume; 2. EEG changes induced by amphetamine, PCP and MK801 are seen in human disease
Outcome Measure	Locomotor activity	-	-	-			In a systematic review, Bahor found that known antipsychotic drugs improved locomotor activity in developmental models of psychosis. In house data from a Masters project 2015 suggests that some clozapine, aripiprazole, fluphenazine but not all eg olanzapine improve cocaine induced locomotor activity.	-	Neither psychomotor agitation nor Cognitive impairment are listed on the JLA schizophrenia PSP top 10, and so the ethological validity of these measures as relevant to unmet clinical need is uncertain
Cognition	-	-	-			We could find no SRs of the effects of known antipsychotic drugs.	-
Additional experimental contrasts	T1A v other	-	-	-		-	In head-to-head experiments, T1A efficacy is non significantly lower than conventional antipsychotics	-	Relevant to potential use as monotherapy
T1A + other v other	-	-	-		-	There is no effect of combined treatment compared with conventional antipsychotic drugs alone	-	Relevant to potential use as component of combination therapy

Dose response relationship: TAAR1 agonists led to significant improvement in locomotor activity, with evidence for a dose-response relationship. There was no significant effect on cognition, and no evidence of a dose-response relationship.

Heterogeneity in observed effects: There was no association between sex; category of disease induction (Genetic or Pharmacological); route of drug administration; whether drug was given before or after disease administration; whether the drug was a full or a partial agonist; whether the drug was highly selective or also had effects at 5-HT receptors; or with study level risks of bias or completeness of reporting, for either of the outcomes considered. This would be unusual for a truly active compound, and suggests, at best, that there have been too few studies adequately to describe the effects of TAAR1 agonists in animal models of schizophrenia.

Small study effects: There was evidence of a significant small study effect for 125 locomotor outcomes, with larger studies giving more smaller estimated for efficacy. There was no significant effect of study size for 19 cognitive outcomes.

Risks of Bias and Reporting Quality: Only 2 of 15 publications were judged at low risk of bias for any of 10 criteria, the majority being at unclear Risk of Bias. There was no significant difference between studies which were of low risk of bias for at least 1 item and those which were not. The median number of ARRIVE reporting guideline items met was 13 out of 22. There was no significant effect of the number of ARRIVE checklist items met, whether this was considered as quartile categories or as a continuous variable.

Balduzzi, Sara, Gerta Rücker, and Guido Schwarzer. 2019. “How to Perform a Meta-Analysis with R: A Practical Tutorial.” Evidence-Based Mental Health, no. 22: 153–60.

Bates, Douglas, Martin Maechler, and Mikael Jagan. 2023. Matrix: Sparse and Dense Matrix Classes and Methods. https://CRAN.R-project.org/package=Matrix.

Bolker, Ben, Gregory R. Warnes, and Thomas Lumley. 2022. gtools: Various r Programming Tools. https://CRAN.R-project.org/package=gtools.

Crippa, Alessio, and Nicola Orsini. 2016. “Multivariate Dose-Response Meta-Analysis: The dosresmeta R Package.” Journal of Statistical Software, Code Snippets 72 (1): 1–15. https://doi.org/10.18637/jss.v072.c01.

Dahl, David B., David Scott, Charles Roosen, Arni Magnusson, and Jonathan Swinton. 2019. xtable: Export Tables to LaTeX or HTML. https://CRAN.R-project.org/package=xtable.

Evans, Robin. 2022. rje: Miscellaneous Useful Functions for Statistics. https://CRAN.R-project.org/package=rje.

Gasparrini, A., B. Armstrong, and M. G. Kenward. 2012. “Multivariate Meta-Analysis for Non-Linear and Other Multi-Parameter Associations.” Statistics in Medicine 31 (29): 3821–39.

Gilbert, Paul, and Ravi Varadhan. 2019. numDeriv: Accurate Numerical Derivatives. https://CRAN.R-project.org/package=numDeriv.

Haddaway, Neal R, Matthew J Page, Chris C Pritchard, and Luke A McGuinness. 2022. “PRISMA2020: An r Package and Shiny App for Producing PRISMA 2020-Compliant Flow Diagrams, with Interactivity for Optimised Digital Transparency and Open Synthesis.” Campbell Systematic Reviews 18 (2): e1230. https://doi.org/10.1002/cl2.1230.

Harrell Jr, Frank E. 2023a. Hmisc: Harrell Miscellaneous. https://CRAN.R-project.org/package=Hmisc.

———. 2023b. rms: Regression Modeling Strategies. https://CRAN.R-project.org/package=rms.

McGuinness, Luke A, and Julian PT Higgins. 2020. “Risk-of-Bias VISualization (Robvis): An r Package and Shiny Web App for Visualizing Risk-of-Bias Assessments.” Research Synthesis Methods. https://doi.org/10.1002/jrsm.1411.

Nakagawa, Shinichi, Malgorzata Lagisz, Rose E. O’Dea, Patrice Pottier, Joanna Rutkowska, Alistair M. Senior, Yefeng Yang, and Daniel W. A. Noble. 2023. “orchaRd 2.0: An r Package for Visualizing Meta-Analyses with Orchard Plots.” EcoEvoRxiv 12: 4–12. https://doi.org/https://doi.org/10.32942/X2QC7K.

Pedersen, Thomas Lin. 2023. patchwork: The Composer of Plots. https://CRAN.R-project.org/package=patchwork.

R Core Team. 2023. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/.

Viechtbauer, Wolfgang. 2010. “Conducting Meta-Analyses in R with the metafor Package.” Journal of Statistical Software 36 (3): 1–48. https://doi.org/10.18637/jss.v036.i03.

White, Thomas, Daniel Noble, Alistair Senior, W. Kyle Hamilton, and Wolfgang Viechtbauer. 2022. metadat: Meta-Analysis Datasets. https://CRAN.R-project.org/package=metadat.

Wickham, Hadley, Mara Averick, Jennifer Bryan, Winston Chang, Lucy D’Agostino McGowan, Romain François, Garrett Grolemund, et al. 2019. “Welcome to the tidyverse.” Journal of Open Source Software 4 (43): 1686. https://doi.org/10.21105/joss.01686.

Wickham, Hadley, Jennifer Bryan, Malcolm Barrett, and Andy Teucher. 2023. usethis: Automate Package and Project Setup. https://CRAN.R-project.org/package=usethis.

Wickham, Hadley, Jim Hester, Winston Chang, and Jennifer Bryan. 2022. devtools: Tools to Make Developing r Packages Easier. https://CRAN.R-project.org/package=devtools.

Xie, Yihui. 2014. “knitr: A Comprehensive Tool for Reproducible Research in R.” In Implementing Reproducible Computational Research, edited by Victoria Stodden, Friedrich Leisch, and Roger D. Peng. Chapman; Hall/CRC.

———. 2015. Dynamic Documents with R and Knitr. 2nd ed. Boca Raton, Florida: Chapman; Hall/CRC. https://yihui.org/knitr/.

———. 2023. knitr: A General-Purpose Package for Dynamic Report Generation in r. https://yihui.org/knitr/.

Zhu, Hao. 2023. kableExtra: Construct Complex Table with “kable” and Pipe Syntax.